Punjab Property Tax Experiment
Pakistan faces important policy challenges in improving service delivery, growth and development. Low levels of tax revenue act as a serious constraint to economic growth, provision of services and, more generally, to building an effective state. Pakistan does poorly on revenue collection, even when compared to other developing countries. To address this problem, the Excise and Taxation (E&T) Department in Punjab in collaboration with CERP designed and implemented a series of human resource reforms beginning in 2009 designed to appropriately incentivize tax collection and improve overall departmental performance. The Property Tax Experiment in Punjab involves the design and evaluation of two separate but complementary HR strategies to increase revenue mobilization without adversely effecting taxpayer satisfaction, stage 1 and stage 2.
The two HR strategies to be designed and evaluated involved performance based pay to incentivize property tax collection staff for stage 1 and merit based transfers and postings for property tax collection staff for stage 2. Stage 1, the study on performance based pay, reached completion in June 2013. The second stage involving merit based transfers and postings is ongoing, now in its second year.
Design and Evaluation
When considering alternative designs for performance based pay or merit based transfers and postings it becomes obvious that there are numerous variations in the form that such policies could assume. How do we find the best possible design for the particular context that we are looking at? Modern development economics has made great strides in design of rigorous evaluation techniques to scientifically gauge the impact of policy interventions. The gold standard of program evaluation is Randomized Control Treatment (RCT) Design.
This involves random selection of some eligible units for program interventions (called treatment) while letting the remaining units (called the control) operate under business as usual. The impact is evaluated by comparing the performance of treatment and control groups over time. The advantage of this approach is that it delivers the true, scientific impact and is not troubled by factors specific to some units and not others. These gold standard results legitimize policy interventions and facilitate scale-up of good policy.
Strategies involving Performance-Based Pay
All 482 tax circles in Punjab were randomized into one of three treatment groups, or a control group. The treatments were designed to measure the trade-offs that the government would pay in terms of increased revenue versus political costs in terms of dissatisfaction among the public. To our knowledge, this is the first RCT in the world that assesses these trade-offs.
The main treatment consisted of three performance-pay schemes introduced for tax-circle staff.
Revenue Based Honorarium Scheme: This scheme motivated tax collectors through the use of output-based incentives in combination with Revenue Based honoraria.
Revenue PLUS Honorarium Scheme: This scheme was designed to be similar in nature to the Revenue Based Honorarium; however checks against over-aggressive tax collection were incorporated by factoring in assessment accuracy and taxpayer satisfaction by utilizing an objective third-party customer feedback assessment conducted on a randomly selected sample of properties in each tax circle.
Conditional Fixed Wage (Flexible Bonus) Scheme: In this scheme all circles are guaranteed a small honorarium but the majority of the honorarium is made at the end of the year and is conditional on performance. In the 2nd year of the study we added an additional control to assess monitoring and salience effects. In general we treat the Information group as a version of a control and include it in the control group.
Based on feedback from the E&T Department, an additional scheme was added for supervisors:
Revenue Based Honorarium for AETOs/ETOs Scheme: The Revenue Based Honorarium Scheme was extended to Assistant Excise and Taxation Officers (AETOs) and Excise and Taxation Officers (ETOs) in the second year of the project.
The incentive schemes produced a substantial and unambiguous positive impact on revenue collection. Treatment circles outperformed control circles by a margin of over 12 percentage points in total collections over the two-year treatment period. Of the three schemes, the Revenue scheme performed best in terms of impact on collections. In both years, the Revenue scheme consistently had the largest effect and the largest return on investment (ROI). Furthermore, a third party survey suggests that the Excise and Taxation Department did not suffer any detectable quality of service costs (either in terms of customer satisfaction or assessment accuracy) as a result of incentivizing inspectors.
We feel the points are noteworthy and should form the basis of a more comprehensive performance-pay system in the E&T department and potentially other similar departments:
- Performance pay works in raising revenues. If revenue increase is an important outcome for the government, some form of monetary incentives has to be an important part of the performance management process for field level staff.
- Simpler and objective performance pay schemes perform better. A key element of an effective performance pay scheme is simple and clear directions that explicitly link to performance on objective dimensions.
- Performance pay schemes may need to be monitored to ensure customer satisfaction. A general concern with performance-pay schemes that only reward on collections is that they may lead to customer dissatisfaction and over-taxation. While our current findings do not show strong evidence for these concerns, it is recommended that the level of customer satisfaction be monitored regularly.
- It may be more Cost-Effective to Introduce Performance-Pay ‘Periods’ Every Few Years. Preliminary evidence suggests that the benefits of performance pay may continue even after the performance-pay period is over; such a persistent effect means that it may be more cost effective for the government to introduce performance-pay schemes every few years.
- Performance Pay Schemes may have to be Designed Differently for Supervisory Tiers. Results of introducing the simplest Revenue scheme (which worked the best for field staff) for supervisory staff were not conclusive; hence further study would be required to design an effective supervisory scheme.
Strategies for Merit Based Transfers and Postings
Merit-based staff allocations have long been the desired goal of human resource policies of most public sector agencies, and the explicit objective of many civil service and tax reforms. There are two important aspects of transfers and postings schemes:
- Allocation (posting inspectors in places where they may have better impact)
- Incentives (using merit-based future transfers to incentivize current performance).
The merit-based transfers and postings schemes will eventually target approximately 240-250 circles in Punjab through ballots. The study was split up into three phases starting with the pilot phase.
Inspectors in circles selected for the pilot will be allocated randomly into two groups of ten and will rank circles in their group in order of preference. The inspectors themselves will be ranked by performance during the previous financial year and will be assigned to circles according to their relative performance and preferences. That is, the first-ranked inspector in a group will be assigned to his most preferred circle in that group, the second-ranked inspector will be assigned to his most preferred circle among remaining circles in that group, and so on.
Both groups of pilot circles will continue in the same scheme in the subsequent year as well. That is, they will again get to choose (in order of performance as noted above) their preferred circle for the FY2014-15 from amongst their group of 10 circles. This choice will be computed in a similar manner as before, and be based on recovery performance in FY 2013-14 as compared to that in FY 2012-13.
In this ballot, 150 circles were selected from across Punjab. Inspectors in these groups will be assessed on their performance during the current fiscal year (FY 2013-14), and at the end of the fiscal year will be allowed to select a circle within their group of ten circles to be posted to in FY1415. Thus this ballot will provide incentives to inspectors to improve their performance in FY2013-14 by allowing them to choose postings for FY2014-15. As in the pilot ballot, all inspectors were allocated randomly into groups of approximately ten circles and will be ranked on their performance throughout the year within their group.
In approximately half of the groups, inspectors will be ranked on the basis of recovery per the evaluative method outlined above for Pilot circles, applied to recovery for FY2013-14, and in the remaining half, they will be ranked by giving weightage to demand. Ranking inspectors by recovery will improve incentives for revenue collection while ranking inspectors by increase in demand will improve incentives to perform well during the department’s survey (when they identify new taxable properties), as inspectors will be motivated to reassess properties thoroughly and discover unregistered properties.
This choice-based transfers’ scheme creates a “tournament” style competition among inspectors, encouraging staff to work harder to compete. We will evaluate the scheme to understand how effective such non-pecuniary incentives are in improving performance and whether, when given choice, inspectors choose circles that they are better matched to in terms of performance. It should be noted that even if inspectors choose circles to maximize their own convenience, it is possible that this by itself can also improve their performance.
In Phase 2, the process will be repeated: half of Phase I circles (selected by ballot), along with additional 70-80 circles selected by ballot that did not participate in Phase I will repeat the same procedure. That is, inspectors in Phase II will once again choose postings in FY 1516 based on FY 1415 performance. In addition, an additional 70-80 circles chosen by ballot will be posted using a central allocation mechanism (such as a ballot procedure). Performance comparisons between these circles will reveal the relative potential of the different schemes and isolate both the incentive and allocative components that comprise an objective transfers and posting system. This will allow us to compare different allocation procedures and determine which are more beneficial in improving staff performance and increasing revenue collection.
Results and policy implications from this second set of strategies based on merit based transfers and postings are forthcoming.
Other Benefits from Engagement
CERP’s engagement with the Excise and Taxation Department at the Government of Punjab has yielded a number of ancillary benefits beyond the design and evaluation of optimal HR strategies.
CERP collated and digitized the past seven years of tax collections data at the circle level staring with FY08/09. All circles provided quarterly circle statements (reports), which were digitized and then verified through the consistency checks mentioned earlier, something that had never been done before at the E&T Department. Combined with personnel data, which was also digitized, this data allows the department ready access to circle and circle-staff level historical performance.
Specifically, the data makes it easy to track changes over time, both in terms of how specific areas and regions are performing in terms of changes in collections, but also in tracking how individual staff members perform over time. Although the digitization was initially meant to gauge the performance of circles for treatment effects, it led to the formation of other highly beneficial methods of assessment. As a result of this exercise, the department is now piloting a new and more detailed reporting format. This format will not only make it easier to understand the yearly workload of circle staff, but it will also provide deeper insight into the variation in cyclical recovery activities, with the eventual goal of moving to a new system of computerized reporting.
Platforms for Collaboration between Researchers and Policy Makers
Including researchers in the policy design process allowed them to serve as a neutral set of experts, fostering more collaborative decision-making within and between various government departments. Within the department, this helped allay concerns of senior leadership as to whether the field staff was diligent in their duties, and concern of the field staff as to whether their interests and constraints were accounted for. Across departments, since the researchers were not associated with any government body, they were able to effectively communicate and at times even advocate one department’s preferences and objectives to another in order to mediate a mutual understanding.
Data Visualization Portal
Finally, in order to fully utilize all of the above beneficial by-products, a ‘proof of concept’ web-based data visualization tool was developed for the department. This tool can facilitate supervisors in the department to monitor circle-wise progress in recovery, track increases in Net-Demand, analyze trends in revenue increase and help develop targeted strategies to improve division-wise property tax collection in an intuitive and visual manner. The tool can generate heat maps based on different variables and show how they vary across circles, districts or divisions. Heat maps also provide patterns of high- and low-collection tax circles, which combined with maps of unassessed areas can help identify localities with a higher marginal return to effort for increasing property tax collection.
Prototypes of these digitized maps are shown below
In this map, the green area shows the current rating area as defined by the Town Municipal Authority (TMA) and Excise and Taxation Department of Punjab. The blue area represents the current Tax Circles of Faisalabad whereas the red area captures properties which are currently not part of the tax circles.
This map shows the performance of all the tax circles in Faisalabad city as measured by the recovery rate against net demand for the year 2011-2012
The tax circles in Sheikhpura city are represented by the blue area and the unassessed area not falling in any tax circle is in red.
This map shows the performance of Tax Circles in Sheikhpura as measured by the recovery rate against net demand for the year 2011-2012